Main
Manushi Welandawe
As a statistician and data scientist, I focus on building fast, robust methods that bridge the gap between theory and real-world data challenges. My work combines scalable Bayesian inference, stochastic optimization, and practical modeling tools to accelerate scientific discovery and decision-making across different fields. I’m passionate about creating solutions that are both statistically sound and computationally efficient.
Experience
Graduate Research Fellow
Boston University
Boston, MA
Present - 2020
- Developing novel diagnostics for detecting stationarity in variational and Bayesian inference contexts
- Developed a robust framework for reliable variational inference with convergence diagnostics and variational approximation assessment
Senior Teaching Fellow
Boston University
Boston, MA
2023
- Instructor for MA 214: Applied Statistics, providing instruction and hands-on support to students; evaluated and graded final group projects to assess applied statistical analysis skills
NSF-MSG Intern
Argonne National Laboratory
Lemont, IL
2022
- Investigated theoretical and empirical properties of gradient estimators in zeroth-order/derivative-free stochastic optimization
Graduate Teaching Fellow
Boston University
Boston, MA
2021 - 2019
- Led 3 discussions each week for graduate level course MA 585 Time series and Forecasting
- Led 5 discussions each week for MA113 Elementary Statistics
- Led 4 discussions each week for MA116 Statistics II
Graduate Researcher
University of Rhode Island
Kingston, RI
2019 - 2018
- Designed a Bayesian mixed-effects zero-inflated beta regression model for longitudinal microbiome data with missing-at-random patterns, validated via simulations and applied to real-world datasets
Graduate Administrative Assistant
University of Rhode Island
Kingston, RI
2019 - 2018
- Organized and conducted workshops on R, SAS, and SPSS for the University of Rhode Island (URI) faculty, graduate students, and undergraduate community
- Provided statistical consultation to local researchers on projects utilizing R, SAS, and SPSS
Graduate Teaching Assistant
University of Rhode Island
Kingston, RI
2017
- Led 5 discussions per week for STA 220 Statistics in Modern Society
- Graded final semester exams for STA 308 Introductory Statistics
Junior Analyst
Peppercube Consultants (Pvt.) Ltd.
Sri Lanka
2017
- Conducted statistical analysis for market research to gain insights in existing or newly developed products and services
Teaching Assistant
University of Sri Jayewardenepura
Sri Lanka
2016
- Led discussions, graded homework, and proctored final semester exams for STA 122/221 Data Analysis I/II
Publications
A framework for improving the reliability of black-box variational inference
Journal of Machine Learning Research
N/A
2024
- Authored with Michael Riis Andersen, Aki Vehtari, and Jonathan H. Huggins.
Challenges and opportunities in high dimensional variational inference
Advances in Neural Information Processing Systems 34 (NeurIPS 2021)
N/A
2021
- Authored with Akash Kumar Dhaka, Alejandro Catalina, Michael R Andersen, Jonathan Huggins, and Aki Vehtari
A Survival Analysis of the Gulf Stream Warm Core Rings
Journal of Geophysical Research: Oceans
N/A
2020
- Authored with E. Nishchitha S. Silva, Avijit Gangopadhyay, Gavin Fay, Glen Gawarkiewicz, Adrienne M. Silver, Mahmud Monim, and Jenifer Clark
Effects of early life NICU stress on the developing gut microbiome
Developmental Psychobiology
N/A
2019
- Authored with Amy L. D’Agata, Jing Wu, Samia V. O. Dutra, Bradley Kane, and Maureen W. Groer
Invited Talks
A Framework to Enhance the Reliability and Detect Convergence of Black-box Variational Inference
N/A
N/A
2024
New England Statistics Symposium
Robust, Automated, and Accurate Black-box Variational Inference
N/A
N/A
2022
Bayes Reading Group, Flatiron Institute